Cosine similarity can force points to grow in magnitude, causing convergence issues.
problem Cosine similarity loss can lead to convergence issues in deep learning.
method Analyzing under-explored settings and proposing cut-initialization.
result Cosine similarity optimization forces points to grow in magnitude, leading to convergence issues.
Adapts auxiliary losses using gradient similarity to improve neural network performance.
problem Statistical inefficiency in neural networks and difficulty in selecting helpful auxiliary tasks.
method Uses cosine similarity between gradients of tasks to adaptively weight auxiliary losses.
result Guaranteed convergence to critical points of the main task and practical usefulness across domains.
Modified cosine distance improves similarity performance in data with variance and correlation.
problem Limitations of traditional cosine similarity in random variable spaces with variance and correlation.
method Proposed a variance-adjusted cosine distance metric to overcome limitations of traditional cosine similarity.
result Modified cosine distance shows 100% test accuracy in KNN model on the Wisconsin Breast Cancer Dataset.
Feedback alignment methods need to be evaluated for accuracy and gradient cosine similarity.
problem Evaluating feedback alignment methods
method Proposed diagnostic evaluation protocol
result Identified silent failures in standard reporting pair
Study evaluates relevance metrics for similarity-based model explanations.
problem Providing understandable explanations for complex model predictions.
method Evaluated three relevance metrics using three tests.
result Cosine similarity of gradients performs best for explanations.
Empirical study on hyperparameters affecting RL generalization.
problem Understanding how hyperparameters affect RL generalization across different domains.
method Empirical analysis of hyperparameters including architecture, regularization, and RL-dependent variables.
result Complex and interdependent relationships between hyperparameters and generalization performance.
This work shows cosine similarity is equivalent to Pearson correlation for word vectors, but not all vectors are suitable for cosine.
problem The use of cosine similarity for semantic textual similarity is often taken for granted, despite its limitations.
method Characterized cases where Pearson correlation is unfit and introduced rank correlation as an alternative.
result Pearson correlation is equivalent to cosine similarity for many word vectors but not all, and rank correlation can improve performance.
Traditionally, multi-layer neural networks use dot product between the output vector of previous layer and the incoming weight vector as the input to activation function. The result of dot product is unbounded, thus increases the risk of large variance. Large variance of neuron makes the model sensitive to the change o…
This paper tackles noise in raw datasets to improve representation learning efficiency.
problem Noise in real-world datasets degrades representation learning quality.
method Proposes denoising Cosine-Similarity (dCS) loss to learn robust representations.
result Empirical results show the dCS loss outperforms baseline objective functions.
CWGD measures gradient diversity weighted by curvature, improving SGD convergence.
problem Gradient noise in high-curvature directions is underestimated by standard methods.
method CWGD weights gradient diversity by the inverse square root of the Hessian.
result CWGD-Cosine reduces optimization error by up to 20% compared to standard cosine annealing.
Beta-SOD detects and corrects noisy object re-identification using cosine similarity and Beta mixtures.
problem Noisy object re-identification in image datasets.
method Reframed Re-ID as a similarity task, using Siamese networks and Beta mixture models.
result Superior performance in noisy conditions compared to state-of-the-art methods.
Proposes an adversarial process using cosine similarity to improve robustness of models.
problem Improving robustness of models by eliminating subsidiary information.
method Adversarial process using cosine similarity to degrade subsidiary model performance.
result Cosine similarity-based adversarial process efficiently degrades subsidiary model performance.
Paper explains contrastive learning using cosine similarity and proposes mitigations for batch size effects.
problem Understanding and improving contrastive learning through batch size effects.
method Unified framework of cosine similarity, theoretical insights, and auxiliary loss.
result Performance improvement in small-batch settings through proposed auxiliary loss.
The paper examines how well node similarities are preserved by random projections in graph embeddings.
problem The preservation of node similarities under random projections in graph embeddings.
method Investigation of dot product and cosine similarity preservation by random projections over graph matrix rows.
result Random projections produce unreliable embeddings for dot product, especially for high-degree nodes.
Person recognition aims at recognizing the same identity across time and space with complicated scenes and similar appearance. In this paper, we propose a novel method to address this task by training a network to obtain robust and representative features. The intuition is that we directly compare and optimize the cosi…
Improved text generation using transferable rewards from related tasks.
problem Non-differentiable task-specific scores limit the use of policy gradient methods in text generation.
method Transferable Reward Learner that uses model-based rewards for sentence-level and phrase-level similarity.
result Improved performance on semantic evaluation measures in image captioning tasks.
New method improves interpretability of text adversarial examples.
problem Generating interpretable adversarial examples in text is challenging.
method Sparse Projected Gradient Descent (SPGD) with directional and sparsity constraints.
result Improves interpretability and likelihood of adversarial examples compared to state-of-the-art methods.
Coherent Multiplex analyzes real-time wavelet coherence among multiple signals.
problem Identifying and visualizing coherence among multiple time series.
method Fast spectral similarity based on cosine similarity metrics of Fourier-transformed signals and sparse time-frequency wavelet coherence.
result Scalable real-time system for low-latency inference and monitoring of inter-signal relationships.
Novel RL-based NPG improves multi-objective NAS efficiency and performance.
problem Discovering optimal neural architectures with multiple conflicting objectives.
method Non-stationary policy gradient with adaptive reward functions and shared model.
result Framework efficiently approximates full Pareto front and achieves superior performance.
Method compares sentences by cosine similarity of vector projections.
problem Measuring semantic similarity of sentences.
method Cosine similarity of vector projections of sentence groups.
result Advantages over existing methods in preserving word order and syntactic connections.
SPEQ improves quantized neural networks by stochastic precision sharing and cosine similarity loss.
problem Improving quantized deep neural networks for edge devices.
method SPEQ combines stochastic precision sharing and cosine similarity loss for knowledge distillation.
result SPEQ outperforms existing methods in various tasks.
CLS measures dataset similarity through decision rule performance.
problem Measuring dataset similarity in machine learning, especially for transfer learning and domain adaptation.
method Cross-Learning Score (CLS) measures similarity through bidirectional generalization performance of decision rules, linking to cosine similarity under canonical linear models.
result CLS effectively measures dataset similarity and transferability, validated on synthetic and real-world datasets.
The study analyzes neural network predictions of knot invariants and finds that braid representations work best.
problem Understanding and predicting knot invariants using neural networks.
method Investigated different knot representations and invariants, proposed a cosine similarity score.
result Braid representations are best for predicting knot invariants, and some invariants are easier to learn than others.
A new topology design improves zero-shot classification performance in contrastive learning.
problem Improving zero-shot classification performance in contrastive visual-textual alignment.
method Proposed an alternative topology design using multiple class tokens and an oblique manifold with negative inner product.
result Improves zero-shot classification performance by an average of 6.1%.
Evaluates SHIELD's effectiveness against adaptive adversaries in various threat models.
problem Evaluating SHIELD's efficacy against adaptive adversaries in different threat models.
method Empirical analysis of SHIELD's robustness against adaptive attacks using Projected Gradient Descent (PGD) attacks in various threat models (white-box, gray-box).
result The targeted PGD attack success rate drops from 64.3% to 48.9% when models are trained from scratch instead of retrained.
We develop a family of techniques to align word embeddings which are derived from different source datasets or created using different mechanisms (e.g., GloVe or word2vec). Our methods are simple and have a closed form to optimally rotate, translate, and scale to minimize root mean squared errors or maximize the averag…
New LSH methods for tensor data improve efficiency and space usage.
problem Efficiency and space usage issues in LSH for tensor data.
method Proposes new LSH methods using CP and TT decompositions for Euclidean and cosine similarity.
result Space-efficient and scalable LSH for tensor data.
The paper analyzes and validates two step size schedules for SGD: exponential and cosine, proving their adaptivity and performance.
problem The variability of SGD performance due to step size choice.
method Analysis and empirical evaluation of exponential and cosine step sizes.
result Exponential and cosine step sizes are adaptive to noise and achieve optimal performance without tuning hyperparameters.
Study evaluates thresholds for removing noise from DNN weights using random matrix theory.
problem Removing noise from deep neural network weights for better approximation.
method Model weights as signal + noise, use random matrix theory to estimate thresholds, evaluate using cosine similarity.
result Proposed threshold estimation method improves approximation quality.
This paper decomposes generalized O'Hara's energies into components.
problem Decomposing generalized O'Hara's energies to understand their components.
method Using an analogue of Doyle-Schramm's cosine formula, the paper derives a decomposition for generalized O'Hara energies.
result Derives a decomposition for generalized O'Hara energies into three components.
New insights show embedding lengths correlate with semantic properties.
problem Contrastive embedding norms ignore embedding magnitudes but correlate with semantic properties.
method Formal theoretical framework and analysis of optimization dynamics.
result Embedding lengths encode semantic information as a byproduct of training.
Improves GANs by enforcing diverse feature learning.
problem GANs can collapse to a single configuration and be unstable.
method Enforces diverse feature learning by penalizing correlated features.
result Enforces diverse features, stabilizes training, and improves image synthesis.
This study investigates self-supervised learning with Wasserstein distance on tree structures.
problem Improving self-supervised learning methods using Wasserstein distance.
method Utilized Tree-Wasserstein distance (TWD) and Jeffrey divergence regularization for training.
result A simple combination of softmax function and Tree-Wasserstein distance outperforms cosine similarity-based methods.
We study rigidity of polyhedral surfaces and the moduli space of polyhedral surfaces using variational principles. Curvature like quantities for polyhedral surfaces are introduced. Many of them are shown to determine the polyhedral metric up to isometry. The action functionals in the variational approaches are derived …
Riemannian stochastic gradient descent converges faster with increasing batch size.
problem Improving convergence rate of Riemannian stochastic gradient descent.
method Theoretical analysis and numerical investigation of increasing batch size effects.
result Riemannian stochastic gradient descent converges faster with increasing batch size.
Study on generalization in gradient-based meta-learning, showing flatter solutions and coherence between adaptation trajectories.
problem Understanding generalization in gradient-based meta-learning.
method Analysis of objective landscapes, experimental demonstration of solution properties, and empirical evidence on coherence between adaptation trajectories.
result Meta-test solutions become flatter, lower in loss, and further away from the meta-train solution as meta-training progresses, even as generalization starts to degrade.
This paper explores sentence vector properties for automatic summarization.
problem Understanding the internal structure and properties of sentence vectors.
method Compositional sentence vector representations using artificial neural networks.
result Cosine similarity correlates with sentence importance and can identify gaps in summaries.
Improved MoE performance through perturbing cosine router.
problem Representation collapse and parameter redundancy in MoE models.
method Least square estimation of cosine router in MoE, followed by noise addition to improve convergence rates.
result Perturbed cosine router leads to polynomial convergence rates for MoE models.
The paper proposes a new method to evaluate LLM agent responses using ECDF clustering.
problem The standard evaluation of LLM agent responses via majority voting obscures response quality and distribution.
method The paper introduces a novel evaluation framework based on ECDF of cosine similarities and clustering of ECDFs using distances and k-medoids algorithm. result ECDF clustering reveals interpretable group structures in LLM responses, offering insights into agent settings.
The study introduces anytime learning schedules for large language models without fixed horizons.
problem Training large language models without knowing the total training horizon.
method Theoretical analysis and weight averaging to create anytime learning schedules.
result Theoretical and empirical evidence shows that weight averaging with simple step sizes can achieve comparable final loss to well-tuned cosine schedules.
WIPS optimizes inner product weights to approximate various similarities.
problem Learning high-quality node representations and accurate similarities.
method Weighted inner product similarity (WIPS) with adjustable weights.
result WIPS can approximate arbitrary general similarities including positive definite and indefinite kernels.
Our work presents extensive empirical evidence that layer rotation, i.e. the evolution across training of the cosine distance between each layer's weight vector and its initialization, constitutes an impressively consistent indicator of generalization performance. In particular, larger cosine distances between final an…
Modified BP attribution methods often ignore later layers' information, leading to misleading explanations.
problem Misleading explanations from modified BP methods ignoring later layers' information.
method Analysis of 9 modified BP methods including Deep Taylor Decomposition, LRP, Excitation BP, PatternAttribution, DeepLIFT, Deconv, RectGrad, Guided BP.
result Only DeepLIFT does not ignore later layers' information, providing a faithful explanation.
New method RECAST improves uncertainty calibration in neural networks.
problem Improving uncertainty estimation in neural networks for better predictions.
method Proposed RECAST method combining cosine annealing, warm restarts, and Stochastic Gradient Langevin Dynamics.
result RECAST offers the best calibrated measure of uncertainty compared to recent methods.
Researchers establish bounds and continuity of decomposed Möbius energies using cosine formula.
problem Estimating the bounds and continuity of decomposed Möbius energies.
method Using the cosine formula to evaluate upper and lower bounds and modulus of continuity of decomposed energies.
result Affirmative answer to the question of estimating decomposed energies using the cosine formula.
New heuristics improve genetic programming's parent selection for classification problems.
problem Improving genetic programming's parent selection for classification tasks.
method Proposed three heuristics inspired by specific classifiers' characteristics, using similarity measures.
result Combination of agreement-based selection and random selection outperforms classical and state-of-the-art schemes.
Cosine schedule is optimal for discrete diffusion models.
problem Choosing the best discretization schedule for diffusion models.
method Optimized using Fisher-Rao geometry.
result Cosine schedule is Fisher-Rao optimal.
A new weighted FDA method improves face recognition accuracy.
problem Equal treatment of all class pairs in FDA leads to suboptimal performance.
method Cosine-weighted and automatically weighted FDA methods are proposed.
result Improved face recognition accuracy through weighted FDA.